Clusterix-Like BigData DBMS
Vadim A. Raikhlin, Roman K. Klassen · Data Science and Engineering · 2020
Abstract Commercial OLAP systems are economically unavailable for organizations with limited financial capabilities. Analytical processing of large amounts of data in these organizations can be accomplished using open-source software systems on a cost-effective cluster platform. Previously created Clusterix-like DBMS using a regular query processing plan is not efficient enough. Therefore, research on such systems was developed with a focus on a full load of processor cores and using the GPU acceleration (systems Clusterix-N, N—from new) up to the development of a system comparable in efficiency to the open-source system Spark, which is currently considered the most promising. The development methodology was based on the constructive system modeling methodology.